Feat: Guardrail support

[IdkwhatImD0ing]

This pull request adds support for input and output guardrails to the agent workflows and OpenAI provider interfaces, enabling validation and control over agent inputs and outputs. It introduces a sample "spaghetti" input guardrail in the tutorial workflow and updates both function signatures and agent instantiation logic to accept and pass guardrail objects throughout the stack.

Guardrail Integration

• Added InputGuardrail and OutputGuardrail parameters to all OpenAI provider agent run methods (run_agent, run_agent_auto_send, run_agent_streamed, run_agent_streamed_auto_send) in src/agentex/lib/adk/providers/_modules/openai.py and src/agentex/lib/core/services/adk/providers/openai.py, including updates to docstrings and argument passing. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]
• Updated agent instantiation logic to correctly forward guardrail objects to the Agent constructor if provided. [1] [2] [3]

Tutorial Workflow Enhancements

• Added a sample input guardrail function check_spaghetti_guardrail and a corresponding TemporalInputGuardrail instance (SPAGHETTI_GUARDRAIL) to the tutorial agent chat workflow, demonstrating how to block or flag user inputs containing the word "spaghetti".
• Updated the agent workflow's event send logic to include the new input guardrail in the call to run_agent_streamed_auto_send, ensuring the guardrail is enforced during agent interaction.

Code Quality and Documentation

• Improved code comments and docstrings to clarify the purpose and usage of guardrail parameters and agent workflow logic. [1] [2] [3] [4] [5] [6]

These changes collectively make it possible to add custom validation logic at both input and output stages of agent workflows, improving safety and reliability.

[smoreinis]

Also, not sure if I'm missing something but it looks like run_agent and run_agent_auto_send in src/agentex/lib/adk/providers/_modules/openai.py still doesn't have input or output guardrails. I believe the plan (from the PR description) is to add these as well to plumb them through to src/agentex/lib/core/services/adk/providers/openai.py, right?